Gökmen-PG (Path Guide)

Gökmen-PG (Path Guide) is an innovative platform with real-time autonomous navigation capability based on deep learning methods, designed for non-human environments or indoor/outdoor settings where GNSS satellite signals are inaccessible. This platform can be used as rotary-wing unmanned aerial vehicles (UAVs) capable of mapping in challenging environmental conditions, as well as automated mobile robots (AMRs) operating in warehouse and logistics management.

Autonomous navigation in an unknown or uncertain environment is one of the challenging tasks for unmanned vehicles. To overcome this challenge, it is necessary to have advanced high-level control methods that are capable of learning and adapting to changing conditions. Until recently, this challenge has been addressed by using hardware such as GPS and Lidar. However, factors such as the absence of satellite signals in challenging environmental conditions or the low efficiency of their simultaneous operation necessitate the creation of new technologies.

Gökmen-PG offers a solution for simultaneous navigation and environmental awareness in unknown environments by eliminating all the disadvantages of GPS and Lidar through a deep learning-based convolutional neural network methodology. This solution is especially anticipated to be used in fields such as search/rescue, security, and autonomous vehicles within the defense industry.


-- SITL Simulations --

-- WebPG --

WebPG is a domestically developed, web-based intelligent mission management platform API (Application Programming Interface). With WebPG, users can watch live video streams, track the movement of aerial vehicles on a map, and remotely plan and control various missions. The platform includes AI models pre-packaged as Docker images, which can perform intelligent inference on raw video data. This allows users to conduct real-time intelligent video analysis without the need to develop their own AI models.

The WebPG API provides data collection and analysis capabilities that can be applied across various fields such as research, agriculture, search and rescue, security, and many more.

-Mechanical Features-

First Iteration 

Designed by Evolutionary Optimization Algorithms 

----------------------------------------------------------------------

Last Iteration

Object Detection YOLOV4-Tiny + TensorRT

ROS - WebSocket Connection

-- Software Architecture --

--- PARTNERS ---